Gut microbiota of breastfed infants are dominated by bifidobacteria( Reference Koropatkin, Cameron and Martens 1 , Reference Harmsen, Wildeboer-Veloo and Raangs 2 ). Bifidobacteria contribute to infant health by preventing certain conditions, such as infections( Reference Fukuda, Toh and Hase 3 – Reference Taipale, Pienihakkinen and Isolauri 5 ) and allergies( Reference Kalliomaki, Kirjavainen and Eerola 6 , Reference Enomoto, Sowa and Nishimori 7 ). The physiological functions of SCFA, such as acetate, produced by bifidobacteria and the direct modulation of the host immune response by bifidobacteria may underlie these preventive effects( Reference Fukuda, Toh and Hase 3 , Reference Mohan, Koebnick and Schildt 8 , Reference de Kivit, Saeland and Kraneveld 9 ). Bifidobacteria populations are often smaller in the gut microbiota of formula-fed infants than in breastfed infants( Reference Koropatkin, Cameron and Martens 1 , Reference Harmsen, Wildeboer-Veloo and Raangs 2 ). Breastfed infants also have decreased infection risks and fewer allergies than formula-fed infants( Reference Hylander, Strobino and Dhanireddy 10 – Reference Saarinen and Kajosaari 14 ). Thus, exclusive breastfeeding is strongly recommended for optimal development of gut microbiota, defence molecules and growth factors( Reference Newburg and Walker 15 , Reference Lonnerdal 16 ). Therefore, an important challenge for the development of infant formula is to provide infants who cannot be breastfed sufficiently with a bifidobacteria-enriched gut microbiota similar to that of breastfed infants.
Non-digestive oligosaccharides, a class of prebiotics, can selectively promote the growth of bifidobacteria by providing an energy source specific for bifidobacteria( Reference Sela, Chapman and Adeuya 17 – Reference Xiao, Takahashi and Nishimoto 19 ). More than 200 types of structurally distinct oligosaccharides have been found in human breast milk( Reference German, Freeman and Lebrilla 18 , Reference Zivkovic, German and Lebrilla 20 ). These human milk oligosaccharides (HMO) are thought to contribute to the development of the bifidobacteria-dominated gut microbiota in breastfed infants( Reference Garrido, Barile and Mills 21 ). The structural diversity of HMO appears to be important to the superior bifidogenic properties of human milk, as oligosaccharides with different structural features require different microbial metabolic processes to be utilised and therefore exhibit different prebiotic properties( Reference Xiao, Takahashi and Nishimoto 19 , Reference Watson, O’Connell Motherway and Schoterman 22 ). Infant formula has traditionally relied on alternative commercially available non-digestive oligosaccharides for their bifidogenic effects, as the diverse and complex HMO are difficult to industrially reproduce. Indeed, there is evidence of prebiotic effects in infant formulas containing commercially available non-digestive oligosaccharides, such as lactulose (LAC)( Reference Kiyosawa, Takase and Yamauchi 23 , Reference Elkington 24 ), raffinose (RAF)( Reference Hattori, Sasai and Yamamoto 25 ), galacto-oligosaccharide (GOS)( Reference Williams, Choe and Price 26 ) and fructo-oligosaccharide( Reference Veereman 27 ). However, the microbiota composition of infants fed these formulas is different from that of breastfed infants( Reference Xia, Williams and Hustead 28 ). Therefore, it is important to clarify and validate the bifidogenic effects of these oligosaccharides on the infant gut microbiota. In vitro fermentation systems using infant faeces have been developed( Reference Satoh, Odamaki and Namura 29 , Reference Rossi, Corradini and Amaretti 30 ) to screen for better oligosaccharides, but studies using human faecal samples are complicated and less reproducible.
Because different oligosaccharides exhibit different prebiotic properties( Reference Xiao, Takahashi and Nishimoto 19 ), we hypothesised that a combination of oligosaccharides would promote bifidobacteria growth in the gut microbiota of infants better than each oligosaccharide alone. This study evaluated the effect of oligosaccharide combinations on the growth of Bifidobacterium breve and Bifidobacterium longum, two major bifidobacteria species found in the gut of breastfed infants( Reference Garrido, Barile and Mills 21 ). We used an in vitro mixed culture model of artificial infant microbiota that was previously reported in a mouse model with humanised microbiota( Reference Martin, Wang and Sprenger 31 ). We also conducted an in vivo experiment using neonatal mice to validate the results of our in vitro experiment. We selected LAC as a basic oligosaccharide because it is one of the most time-proven prebiotics and has been used in commercial infant formula products for about 50 years( Reference Elkington 24 ). We assessed the combination of LAC with RAF and/or GOS.
Methods
Oligosaccharides
LAC (MLC-97, 99·9 %; Morinaga Milk Industry Co. Ltd), RAF (Nitten Raffinose, >98 %; Nippon Beet Sugar Manufacturing Co. Ltd) and GOS (composed of approximately 14 % tetrasaccharides, 82 % trisaccharides and 2 % disaccharides) were used in this study. GOS was obtained by purification of Oligomate 55N (Yakult Honsha Co. Ltd) using gel filtration column chromatography, as previously reported( Reference Satoh, Odamaki and Namura 29 ).
Bacterial strains
Bacterial strains were obtained from the American Type Culture Collection (ATCC) and the Japan Collection of Microorganisms (JCM). Seven bacterial species were used to create the artificial human baby flora culture( Reference Martin, Wang and Sprenger 31 ): B. breve (M-16V; Morinaga Milk Industry)( Reference Abe, Muto and Yaeshima 32 ), B. longum (JCM1217T), Escherichia coli (JCM1649T), Staphylococcus aureus (JCM20624T), Staphylococcus epidermidis (JCM2414T), Clostridium perfringens (ATCC13124T) and Parabacteroides distasonis (JCM5825T).
In vitro fermentation experiment
The seven bacterial strains were cultured individually at 37°C in Gifu Anaerobic Medium broth (Nissui Pharmaceutical Co. Ltd) to inoculate the test medium. Optical densities of cultures were monitored and bacteria were collected during growth phase by centrifugation. All strains were resuspended in a complex medium (FM medium)( Reference Satoh, Odamaki and Namura 29 , Reference Rossi, Corradini and Amaretti 30 ) containing yeast extract (5 g/l), ascorbic acid (10 g/l), sodium acetate (10 g/l), (NH4)2SO4 (5 g/l), urea (2 g/l), MgSO4.7H2O (0·2 g/l), FeSO4.H2O (0·01 g/l), MnSO4.H2O (0·007 g/l), NaCl (0·01 g/l), Tween 80 (1 g/l), hemin (0·05 g/l) and l-cysteine hydrochloride monohydrate (0·5 g/l), and 2×107 cells/ml of each strain were then pooled together. This bacterial mixture was used as a seed culture. The in vitro fermentation was performed according to a previous report( Reference Satoh, Odamaki and Namura 29 ) using a pH-controlled, multi-channel fermenter (Bio Jr.8; ABLE). In brief, the test medium was prepared as follows: 100 ml of FM medium containing either LAC, LAC–RAF, LAC–GOS or LAC–RAF–GOS (total oligosaccharide concentration of 1 % w/v) was incubated with CO2 gas overnight to achieve anaerobic conditions at a pH of 7·0. The ratios of oligosaccharides in the culture medium were 1:1 and 1:1:1 for two- and three-oligosaccharide combinations, respectively. The test medium was inoculated with 1 ml of the seed culture to create the artificial human baby flora, resulting in test medium containing 2×105 cells of each strain. After inoculation, fermentation was carried out at 37°C with continuous CO2 flow, gentle stirring and no pH regulation. After a 24-h fermentation, 1 ml of the test culture was sampled to measure bacterial cell numbers and SCFA.
Measurement of bacterial cell numbers
Bacterial cell numbers in the culture and in the large intestine content of mice were determined by a real-time PCR method using primers specific to the 16S rRNA gene of each bacterial group, as previously reported( Reference Satoh, Odamaki and Namura 29 ). Bacterial pellets from 1 ml of the mixed culture, or approximately 20 mg of the large intestinal content from neonatal mice, were vigorously homogenised in 500 μl of extraction buffer (100 mm-Tris-HCl, 40 mm-EDTA and 1 % of SDS at pH 9·0) with 300 mg of 0·1-mm-diameter glass beads and 500 μl of TE (Tris-EDTA; 10 mM-Tris-HCl, 1 mM-EDTA, pH 8·0)-saturated phenol for 30 s using the FastPrep FP100A bead cell disruptor (Qbiogene Inc.) at a power level of 5·0. Bacterial DNA was extracted and purified using a standard phenol–chloroform method. Purified DNA was subjected to real-time PCR using the 7500 Fast Real-Time PCR System (Life Technologies) and SYBR Premix Ex Taq (Takara Bio Inc.) with primers specific to the 16S rRNA gene of each bacterial group to determine the cell numbers of each bacterial strain in the samples. Standard DNA for each bacterial strain was also prepared from a pure culture of each strain, in which the bacterial cell number was microscopically counted. These standard DNA were subjected to real-time PCR together with samples to construct standard curves for the determination of bacterial cell numbers in the samples. Detailed primer information is summarised in the online Supplementary Table S1.
Measurement of SCFA
SCFA were measured as previously described( Reference Satoh, Odamaki and Namura 29 ). In brief, culture supernatants were diluted 5-fold with 3·75 mm-perchloric acid and then ultrafiltrated using a NANOCEP 10K Omega spin column (Pall Corporation). Caecum contents from neonatal mice were homogenised in ten volumes of distilled water and then centrifuged at 13 000 g for 10 min at 4°C. Nine volumes of the resulting supernatant were mixed with one volume of 30 mm-perchloric acid and then ultrafiltrated as described above. The concentrations of SCFA in ultrafiltrated samples were analysed using an ion-exclusion HPLC system (Shimadzu)( Reference Satoh, Odamaki and Namura 29 ). Concentrations of acetate in the culture supernatants were determined by subtracting the initial acetate concentration of FM medium from the acetate concentrations after fermentation.
Measurement of lactulose
Culture supernatants were ultrafiltrated using a NANOCEP 10K Omega spin column. Oligosaccharides in the ultrafiltrated samples were separated by an ion-exclusion HPLC with the following conditions: column, a TSKgel SUGAR AXI column (4·6×150 mm) (Tosoh); column temperature, 60°C; flow rate, 0·4 ml/min; mobile phase, 0·5 m-borate buffer, pH 8·7. Then, separated LAC was derivatised with arginine and detected using the post-column fluorometric detection method( Reference Hardy, Townsend and Lee 33 ) with a RF-20Axs fluorescence detector (Shimadzu). Consumption rates of LAC in the culture supernatants were determined by subtracting the concentrations of LAC remaining after the 24-h fermentation from the initial concentrations.
Animal studies
All animal experiments were approved by and performed in accordance with the guidelines of the Animal Research Committee of Morinaga Milk Industry Co. Ltd. Lactating female C57BL6 mice aged 10 weeks old with 2-d-old (D2) male pups were obtained from Japan SLC Inc. The litter size was adjusted to four. Mothers were fed ad libitum with a standard rodent diet (F-2; Funabashi Farm) and pups were nursed by their mothers throughout the study. The mice were exposed to a 12 h light–12 h dark cycle and the room was maintained at a constant temperature of 23 (sem 2)°C. To evaluate the effects of the oligosaccharide combinations on the growth of bifidobacteria, pups were divided into five groups and orally administered with 100 μl of PBS containing one of the following five mixtures once a day from D6 to D14: vehicle (PBS alone), 5×107 B. breve, 20 mg LAC with 5×107 B. breve, 20 mg LAC–RAF mixture with 5×107 B. breve and 20 mg LAC–RAF–GOS mixture with 5×107 B. breve. The w/v ratio of oligosaccharides was 1:1 and 1:1:1 for the LAC–RAF and LAC–RAF–GOS groups, respectively. Pups were killed at D15 by deep anaesthesia with sevoflurane to harvest the large intestine and caecum contents. The detailed protocol of the experiment is illustrated in Fig. 2(A).
Quantitative real-time PCR analysis
Total RNA in the liver and large intestine were extracted using TRIzol reagent (Life Technologies). The quality of the purified total RNA was confirmed by electrophoretic analysis using an Experion Automated Electrophoresis System (Bio-Rad Laboratories Inc.). Complementary DNA was synthesised from 1 μg of total RNA using the QuantiTect Reverse Transcription Kit (Qiagen). Gene expression levels were determined by real-time PCR using the 7500 Fast Real-Time PCR System (Life Technologies) with SYBR Premix Ex Taq (Takara Bio). The specificity of each PCR was confirmed by analysing the melting curve of the amplicon according to the manufacturer’s protocols. All samples amplified by real-time PCR were tested in duplicate. Relative mRNA levels of target genes were determined on the basis of the mRNA levels of the housekeeping gene β-actin (Actb). Detailed primer information is summarised in the online Supplementary Table S2.
Statistical analysis
Data are expressed as means with their standard errors. Differences in bacterial cell numbers, pH of each time point and concentrations of SCFA among the four experimental groups in the in vitro experiment and differences in bacterial cell numbers, concentrations of SCFA and mRNA levels among the five experimental groups in the in vivo experiments were all assessed by the Tukey–Kramer honestly significant difference (HSD) test for multiple comparisons. Data were analysed using JMP software version 5.1 (SAS institute Inc.). All results with P<0·05 were considered statistically significant.
Results
In vitro bifidogenic effects of lactulose and oligosaccharide combinations
We analysed the bifidogenic effects of LAC alone and various combinations of oligosaccharides using a simplified in vitro mixed culture model of artificial infant microbiota. After a 24-h fermentation, the cell numbers of B. breve in the cultures with the LAC–GOS combination and with LAC–RAF–GOS combination were significantly higher than those in the cultures with LAC or LAC–RAF (P<0·05) (Fig. 1(A)). The cell numbers of B. longum were significantly higher in the cultures with LAC–RAF–GOS compared with those in other tested cultures (P<0·05) (Fig. 1(A)). Rapid decreases in pH were observed in the cultures with LAC–GOS and with LAC–RAF–GOS (Fig. 1(B)). Consistent with the cell number of bifidobacteria and the decrease in pH in the cultures after a 24-h fermentation, the concentrations of LAC were decreased by 19·1 (sem 1·0) % in the LAC culture and by 54·2 (sem 6·2) % in the LAC–RAF culture, respectively. In contrast, LAC was not detected in the LAC–GOS and LAC–RAF–GOS cultures after the 24-h fermentation. Consistent with this, concentrations of acetate, a major SCFA produced by the fermentation of carbohydrates by bifidobacteria, were significantly higher in the cultures with LAC–GOS and LAC–RAF–GOS than in the cultures with LAC and LAC–RAF (P<0·05) (Fig. 1(C)), indicating active utilisation of the LAC–GOS and LAC–RAF–GOS combinations. The cell numbers of Staphylococcus were relatively high in the cultures containing LAC–RAF, the cell numbers of E. coli were high in LAC–GOS and LAC–RAF–GOS, the cell numbers of C. perfringens were high in LAC and the cell numbers of P. distasonis were high in LAC–RAF (Fig. 1(A)).
Fig. 1 Combinational effects of oligosaccharides on the bifidobacterial growth in vitro. (A) Bacterial cell numbers of each species in the culture after the 24-h fermentation. (B) pH changes of the culture medium during fermentation. (C) Concentrations of lactate and SCFA in the culture medium after the 24-h fermentation. Values are means (n 4), with their standard errors. a,b,c,d Mean values with unlike letters were significantly different (P<0·05, Tukey–Kramer HSD test). 
, lactulose (LAC); 
, LAC–raffinose (RAF); 
, LAC–galacto-oligosaccharide (GOS); 
, LAC–RAF–GOS; 
, LAC; 
, LAC–RAF; 
, LAC–GOS; 
, LAC–RAF–GOS.
In vivo bifidogenic effects of lactulose, lactulose–raffinose and lactulose–raffinose–galacto-oligosaccharide
We conducted an in vivo experiment to validate the results of our in vitro culture model. We selected three oligosaccharide groups, LAC, LAC–RAF and LAC–RAF–GOS, that appeared to differ in their bifidogenic effects according to the in vitro model for further in vivo analysis.
In the in vivo experiment, neonatal mice were orally administered with one of the oligosaccharide mixtures and B. breve from D6 to D14 (Fig. 2(A)). The cell numbers of B. breve in the large intestine content were measured after a 9-d administration by real-time PCR. 16S rRNA gene copy numbers of total bacteria were slightly higher in the B. breve-administered groups than in the vehicle group (Fig. 2(B)). B. breve were not detected in the vehicle group, but they were detected in all groups administered with B. breve at about 1010 cells/g of faeces (Fig. 2(B)). Levels of B. breve 16S rRNA gene in the gut reflected the trend observed in the in vitro experiment depicted in Fig. 1. The levels of B. breve 16S rRNA gene were relatively low in the LAC group (accounting for 0·8 % of all bacterial 16S rRNA gene, a similar level to the B. breve alone group), moderate in the LAC–RAF group (1·4 %) and high in the LAC–RAF–GOS group (6·0 %) (Fig. 2(B)). The amount of acetate in the caecum was significantly higher in the LAC–RAF–GOS group than in other groups (P<0·05) (Fig. 2(C)). We also analysed the microbiota of these neonatal mice using real-time PCR with primers specific to the 16S rRNA gene of each bacterial group. Compared with the vehicle group, the cell numbers of Enterobacteriaceae were higher in the B. breve alone, LAC and LAC–RAF groups, but remained unchanged in the LAC–RAF–GOS group (Fig. 2(D)). On the other hand, the cell numbers of Bacteroidetes were greater in the B. breve alone and LAC–RAF–GOS groups (P<0·05) (Fig. 2(D)). Firmicutes were slightly greater in the B. breve alone group and were unchanged in the co-administered groups (P<0·05) (Fig. 2(D)). The cell numbers of Lactobacillus were comparable in all groups (Fig. 2(D)).
Fig. 2 Combinational effects of oligosaccharides on the bifidobacterial growth and gut microbiota in neonatal mice. (A) The protocol of the oral administration with oligosaccharides to neonatal mice. (B) 16S rRNA gene copy numbers of total bacteria (left panel), cell numbers of Bifidobacterium breve (middle panel) and 16S rRNA gene copy numbers of B. breve relative to those of total bacteria (right panel) in the large intestinal content of neonatal mice. Data were expressed as per gram wet weight of the large intestinal contents. (C) The amounts of lactate and acetate in the caecum of neonatal mice. (D) Cell numbers of Enterobacteriaceae, Bacteroidetes, Firmicutes and Lactobacillus in the large intestinal content of neonatal mice. Values are means (n 7–8), with their standard errors. a,b,c,d Mean values with unlike letters were significantly different (P<0·05, Tukey–Kramer HSD test). ND, not detected; LAC, lactulose; RAF, raffinose; GOS, galacto-oligosaccharide.
In the in vivo experiment, serum acetate concentrations were increased in the LAC–RAF–GOS group (Fig. 3(A)). We investigated mRNA levels of metabolism- and immune response-related genes in the liver and large intestine by quantitative real-time PCR. We found that the mRNA levels of gluconeogenesis- and fatty acid β-oxidation-related genes were unchanged in the liver of the LAC–RAF–GOS group, except for a slight increase in expression of the β-oxidation enzyme gene, Acadm, relative to other groups (P<0·05) (Fig. 3(B)). On the other hand, mRNA levels of certain markers of helper T-cell (Th) subsets, such as Gata3 (marker of Th2) and Foxp3 (marker of regulatory T cell), but not T-bet (marker of Th1) or Rorc (marker of Th17), were increased in the large intestine of the LAC–RAF–GOS group relative to all other groups (P<0·05) (Fig. 3(C)). In addition, levels of the innate immune sensor gene Tlr2, but not Tlr4, tended to be decreased and levels of the hormone gene Gcg, which produces glucagon-like peptide (GLP)-1 and GLP-2 in the intestine, but not Pyy, were increased in the LAC–RAF–GOS group relative to all other groups (P<0·05) (Fig. 3(C)).
Fig. 3 Combinational effects of oligosaccharides on the serum acetate and gene expression in neonatal mice. (A) Serum acetate concentration. (B) Expression of genes related to gluconeogenesis (G6pc and Pck1) and fatty acid β-oxidation (Acadm and Acadl) in the liver. (C) Gene expression of helper T cell (Th) subset markers (T-bet, Gata3, Rorc and Foxp3), sensors of microbial components (Tlr2 and Tlr4) and gut-derived hormones (Gcg and Pyy) in the large intestine. Values are means (n 7–8), with their standard errors. a,b,c Mean values with unlike letters were significantly different (P<0·05, Tukey–Kramer HSD test). LAC, lactulose; RAF, raffinose; GOS, galacto-oligosaccharide; G6pc, glucose-6-phosphatase, catalytic; Pck1, phosphoenolpyruvate carboxykinase 1, cytosolic; Acadm, acyl-coenzyme A dehydrogenase, medium chain; Acadl, acyl-coenzyme A dehydrogenase, long-chain; T-bet, T-box 21; Gata3, GATA binding protein 3; Rorc, RAR-related orphan receptor gamma; Foxp3, forkhead box P3; Tlr2, Toll-like receptor 2; Tlr4, Toll-like receptor 4; Gcg, glucagon; Pyy, peptide YY.
Discussion
It is important to provide formula-fed infants with a bifidobacteria-enriched gut microbiota similar to that of breastfed infants. Various types of non-digestive oligosaccharides are used in infant formula worldwide to improve the gut microbiota of formula-fed infants, but a comparative validation of these oligosaccharides remains insufficient. In this study, we used an in vitro mixed culture model of artificial infant microbiota with seven bacterial species as a simple method to assess the bifidogenic effects of oligosaccharides alone and in combination. We found that the combination of LAC, RAF and GOS exhibited synergistic effects on the in vitro growth of B. breve and B. longum compared with LAC alone and LAC–RAF and LAC–GOS combinations. This synergistic bifidogenic effect was confirmed in the in vivo experiment using neonatal mice. Our results suggest that the LAC–RAF–GOS combination may be useful in infant formula to promote healthy gut microbiota in formula-fed infants.
Our in vitro mixed culture model of artificial infant microbiota well reflected the in vivo bifidogenic effect of oligosaccharides. This in vitro model may be useful to screen better oligosaccharides for infant. On the other hand, the in vitro model had some experimental limitations. For example, the in vitro model is unable to reproduce the indirect effects of oligosaccharides brought about by changes within the gut microbiota and metabolites, as well as by host–microbe interactions. In fact, cell numbers of Enterobacteriaceae (represented by E. coli in our culture model) were slightly increased in the LAC–RAF–GOS group relative to the LAC and LAC–RAF groups in the in vitro model, but were lower in the neonatal mice administered with LAC–RAF–GOS than in the mice administered with LAC and LAC–RAF. Therefore, our in vitro model is valid as a simple method to evaluate the direct bifidogenic effects of oligosaccharides and can serve as an alternative to faecal culture experiments for this purpose, but it may not be suitable for predicting overall changes in the infant gut microbiota caused by oligosaccharides.
We found that RAF and GOS exhibited different effects on the growth of bifidobacteria in combination with LAC in our in vitro model. LAC–RAF enhanced the growth of B. longum, whereas LAC–GOS enhanced the growth of B. breve, acetate production and pH reduction compared with LAC alone. This is consistent with the results of a previous report that B. breve highly utilised GOS compared with B. longum ( Reference Barboza, Sela and Pirim 34 ). This is also consistent with a recent study showing that B. breve was effectively increased in the gut microbiota of infants fed formula containing GOS as the only prebiotic( Reference Sierra, Bernal and Blasco 35 ). Interestingly, the features of LAC–RAF and LAC–GOS combinations remained or were emphasised in the LAC–RAF–GOS combination, which was high in both B. breve and B. longum cell numbers, high in acetate concentration and rapidly decreased in pH. This was true even though the total oligosaccharide concentration was equal in all media (i.e. the initial concentration of each oligosaccharide was relatively low in the medium with three oligosaccharides compared with the medium with two oligosaccharides). The underlying mechanisms of this synergistic effect remain unknown, but secondary metabolites and inter-species metabolite transfers, as well as increased choices of carbohydrate sources, may be involved. Metabolomic and metatranscriptomic approaches could help understand the overall metabolic process of LAC–RAF–GOS in the in vitro model. Previous in vitro and animal studies reported that structurally different oligosaccharides exhibited different prebiotic effects and fermentation properties( Reference Xiao, Takahashi and Nishimoto 19 , Reference Van Craeyveld, Swennen and Dornez 36 ). Over 200 structurally different oligosaccharides have been identified in human milk( Reference Zivkovic, German and Lebrilla 20 ). Taken together, these reports and our results suggest that an increased structural diversity of oligosaccharide combinations could enhance the bifidogenic effects of infant formula. The LAC–RAF–GOS combination also increased the cell numbers of Bacteroidetes in our in vivo experiment. Recent studies on the composition of the infant gut microbiota using high-throughput 16S rRNA gene sequencing suggest that breastfed infants have more Bacteroidetes than formula-fed infants( Reference Fan, Huo and Li 37 , Reference Donovan, Wang and Li 38 ). Although results from our in vivo experiment must be carefully interpreted because of the differences in gut microbiota between neonatal mice and human infant (dominated by Enterobacteriaceae and containing few bifidobacteria in neonatal mice)( Reference Mirpuri, Raetz and Sturge 39 , Reference Gomez-Gallego, Collado and Perez 40 ), formula with a LAC–RAF–GOS combination could contribute to the development of gut microbiota in formula-fed infants similar to that in breastfed infants by increasing bifidobacteria and Bacteroidetes.
In our in vivo experiment, we also observed that oral administration of the three-oligosaccharide combination increased acetate concentrations in the caecum and serum of neonatal mice. Expression of the fatty acid β-oxidation gene, Acadm, was increased in the liver, and expression of some Th subset markers, including Foxp3 and proglucagon gene Gcg, were increased in the large intestine, suggesting an altered metabolism and immune response in the neonatal mice administered with the three-oligosaccharide combination. With regard to the physiological significance of these observations, there are reports indicating that SCFA, such as acetate, produced by the gut microbiota can prevent the development of host metabolic disorders by several mechanisms, such as by elevating hepatic fatty acid oxidation and intestinal Gcg expression/GLP-1 secretion( Reference Kondo, Kishi and Fushimi 41 – Reference den Besten, van Eunen and Groen 43 ). In addition, recent studies indicate that gut microbiota-derived SCFA can suppress inflammatory and allergic responses by inducing the differentiation of colonic regulatory T cells( Reference Furusawa, Obata and Fukuda 44 ) or by enhancing the cell numbers and function of lung regulatory T cells( Reference Thorburn, McKenzie and Shen 45 ) via epigenetic modification and induction of the Foxp3 gene( Reference Furusawa, Obata and Fukuda 44 , Reference Thorburn, McKenzie and Shen 45 ). Some reports suggest that formula feeding increases the risk of obesity in childhood and later in life( Reference Yamakawa, Yorifuji and Inoue 46 , Reference Oddy, Mori and Huang 47 ) and also increases risks of infection( Reference Hylander, Strobino and Dhanireddy 10 – Reference Sinha, Madden and Ross-Degnan 13 ) and allergies( Reference Saarinen and Kajosaari 14 ) relative to exclusive breastfeeding. Considering the apparent differences between the gut microbiota of breastfed and formula-fed infants, correcting the gut microbiota with the three-oligosaccharide combination could improve host metabolism and immune responses of formula-fed infants. Interestingly, the disruption of the gut microbiota at the neonatal stage results in later metabolic diseases( Reference Cox, Yamanishi and Sohn 48 ), and early-life exposure to acetate prevents later development of allergic airways diseases( Reference Thorburn, McKenzie and Shen 45 ) in animal models. Further studies using pathological models are needed to clarify the effects of intervention in the infant gut microbiota with prebiotic oligosaccharides on long-lasting health.
In conclusion, this study reveals a synergistic bifidogenic effect of the three-oligosaccharide combination, both in vitro and in vivo, providing valuable information for the development of infant formula, as well as for management of the gut microbiota and health of formula-fed infants.
Acknowledgements
The authors thank Dr Xiao, Dr Odamaki, Dr Sato and Sugahara of the Morinaga Milk Industry Co. Ltd for technical support in the in vitro fermentation experiment and in the gut microbiota analysis, and thank Hatano and Tsutsumi of the Morinaga Milk Industry Co. Ltd for technical support in the analysis of lactulose.
T. E., H. I., K. N. and Y. T. designed the study. T. E., H. I. and M. T. performed the experiments and T. E. analysed the data. T. E., H. I., M. T., H. I., K. N. and Y. T. wrote the manuscript. This work was funded by Morinaga Milk Industry Co., Ltd. All the authors read and approved the final manuscript. Morinaga Milk Industry Co. Ltd had no role in the design, analysis or writing of this article.
All authors are employees of, and obtain salaries from, Morinaga Milk Industry Co. Ltd Morinaga Milk Industry Co. Ltd provided the B. breve M-16V and lactulose used in this study.
Supplementary Material
For supplementary material/s referred to in this article, please visit http://dx.doi.org/doi:10.1017/S0007114516001987
